Post Go back to editing

LTM4609 sudden death issues - any possible causes?

We have built a number of boards using the LTM4609 to create a supply for a single board computer 19V ~ 1A.

The idea is that mostly the supply is an external 28V, but when that fails, a series pair of Lithium cells take over, via a diode or, so the input voltage is either 24-32V normally or between 6 and 8V on internal batteries.

When the devices have failed, measuring  an un-powered board, there is a a low impedance from the inductor to ground, while on a healthy module this is very high impedance. I am struggling to see what could be the issue, our current limit resistor  is 15 milliohms, and we do not expect to be anywhere near needing all the current that would alow .  And of course on the Eval module it did not do this, it worked perfectly.

Waveforms do not show any instability and there is no significant kickback on diode or-ed the supply  Vin on switch off.

Any pointers gratefully received at this stage- any particular pin voltages we need to be looking out for or corner case conditions  that make it vulnerable ?

  • Hi mikep-j, 

    Some things that can cause failure for the LTM4609 are listed below.

    1. Inductor current exceeding its saturation current rating (especially if hard-saturating ferrite)

    2. Layout issues such as :

        - Not kelvin sensing directly across the current sense resistor with Sense+/Sense-

        - Not having a sufficiently low impedance path for the high current coming out of the Rsense bank and through the sense resistor to PGND.

          Use plenty of vias if sense resistor is not on the same layer next to the module (ie on opposite layer).

    3. Current sense resistor itself and/or layout has significant equivalent series inductance. Please use low ESL type & kelvin sensing. 

    4. Loop stability margins insufficient (ie unstable switching). 

    High peak inductor currents can cause excessive internal mosfet stress, which eventually can cause mosfets to fail short if not limited properly by the current sense loop. The current sensing signal must be kept clean. The current through the resistor is high and slew rate is fast so requires a low ESL sensing resistor and good kelvin sense routing for it to the sense pins as shown in the datasheet. If not, noise spikes in the signal can cause the current comparator to trip at an improper time leading to high peak current stress conditions in the module. 

    For your application, you may also want to reach out to your local Field Applications Engineer / Sales to make sure you have the best module  solution to date for your power with battery backup requirements. 

    Thanks